git.rot13.org Git - koha-bibliografija/blob - html.pl

   1 #!/usr/bin/perl
   2
   3 # LC_COLLATE=hr_HR.utf8 KOHA_CONF=/etc/koha/sites/ffzg/koha-conf.xml ./html.pl
   4
   5 use warnings;
   6 use strict;
   7
   8 use DBI;
   9 use Data::Dump qw(dump);
  10 use autodie;
  11 use locale;
  12 use Text::Unaccent;
  13 use Carp qw(confess);
  14 use utf8;
  15 use JSON;
  16 use POSIX qw(strftime);
  17
  18 use lib '/srv/koha_ffzg';
  19 use C4::Context;
  20 use XML::LibXML;
  21 use XML::LibXSLT;
  22
  23 my $dbh = C4::Context->dbh;
  24
  25 sub debug {
  26         my ($title, $data) = @_;
  27         print "# $title ",dump($data), $/ if $ENV{DEBUG};
  28 }
  29
  30 my $xslfilename = 'compact.xsl';
  31
  32 my $azvo_group_title = {
  33 'znanstveno nastavni' => qr/(profes|docent|znanstveni savjetnik|znanstveni suradnik)/i,
  34 'lektori i predavači' => qr/(lektor|predavač)/i,
  35 'asistenti i novaci' => qr/(asistent|novak)/i,
  36 };
  37
  38 my $department_groups = {
  39 'AAB_humanističke'             => qr/(anglistiku|arheologiju|antropologiju|filozofiju|fonetiku|germanistiku|hungarologiju|indologiju|slavenske|filologiju|komparativnu|kroatistiku|lingvistiku|povijest|romanistiku|talijanistiku)/i,
  40 'AAC_društvene'                        => qr/(informacijske|pedagogiju|psihologiju|sociologiju)/i,
  41 };
  42
  43 my $auth_header;
  44 my $auth_department;
  45 my $auth_group;
  46 my @authors;
  47 my $department_in_sum;
  48 my $department_in_group;
  49
  50 my $skip;
  51
  52 my $sth_auth = $dbh->prepare(q{
  53 select
  54         authid,
  55         ExtractValue(marcxml,'//datafield[@tag="100"]/subfield[@code="a"]') as full_name,
  56         ExtractValue(marcxml,'//datafield[@tag="680"]/subfield[@code="a"]') as department,
  57         ExtractValue(marcxml,'//datafield[@tag="680"]/subfield[@code="i"]') as academic_title
  58 from auth_header
  59 });
  60
  61 $sth_auth->execute();
  62 while( my $row = $sth_auth->fetchrow_hashref ) {
  63         if ( $row->{department} !~ m/Filozofski fakultet u Zagrebu/ ) {
  64                 push @{ $skip->{nije_ffzg} }, $row;
  65                 next;
  66         }
  67         $auth_header->{ $row->{authid} } = $row->{full_name};
  68         $row->{department} =~ s/, Filozofski fakultet u Zagrebu.*$//;
  69         $row->{department} =~ s/^.+\.\s*//;
  70         $row->{department} =~ s/\s+$//s;
  71         my $group;
  72         foreach my $title ( keys %$azvo_group_title ) {
  73                 if ( $row->{academic_title} =~ $azvo_group_title->{$title} ) {
  74                         $group = $title;
  75                         last;
  76                 }
  77         }
  78         if ( $group ) {
  79                 $row->{academic_group} = $group;
  80                 $auth_group->{ $row->{authid} } = $group;
  81                 $skip->{group_stat}->{$group}++;
  82         } else {
  83                 push @{ $skip->{no_academic_group} }, $row;
  84         }
  85
  86 #       warn "# ", dump( $row );
  87         push @{ $auth_department->{ $row->{department} } }, $row->{authid};
  88         push @authors, $row;
  89         $department_in_sum->{ $row->{department} }++;
  90         foreach my $name ( keys %$department_groups ) {
  91                 my $regex = $department_groups->{$name};
  92                 if ( $row->{department} =~ $regex ) {
  93                         $department_in_group->{ $row->{department} } = $name;
  94                         last;
  95                 }
  96         }
  97 }
  98
  99 debug 'department_in_group' => $department_in_group;
 100
 101 foreach my $department ( keys %$department_in_sum ) {
 102         $department_in_sum->{$department} = 0 unless $department =~ m/(centar|croaticum|katedra|odsjek)/i;
 103 }
 104
 105 debug 'auth_department' => $auth_department;
 106 debug 'auth_group' => $auth_group;
 107 debug 'department_in_sum' => $department_in_sum;
 108
 109
 110 my $authors;
 111 my $marcxml;
 112
 113 my $sth_select_authors  = $dbh->prepare(q{
 114 select
 115         biblionumber,
 116         itemtype,
 117         marcxml
 118 from biblioitems
 119 where
 120         agerestriction > 0
 121 });
 122
 123 =for sql
 124 --      ExtractValue(marcxml,'//datafield[@tag="100"]/subfield[@code="9"]') as first_author,
 125 --      ExtractValue(marcxml,'//datafield[@tag="700"]/subfield[@code="9"]') as other_authors,
 126 --      ExtractValue(marcxml,'//datafield[@tag="942"]/subfield[@code="t"]') as category,
 127
 128 --      and SUBSTR(ExtractValue(marcxml,'//controlfield[@tag="008"]'),8,4) between 2008 and 2013
 129 -- order by SUBSTR(ExtractValue(marcxml,'//controlfield[@tag="008"]'),8,4) desc
 130 =cut
 131
 132 my $biblio_year;
 133 my $type_stats;
 134
 135 my $parser = XML::LibXML->new();
 136 $parser->recover_silently(0); # don't die when you find &, >, etc
 137 my $style_doc = $parser->parse_file($xslfilename);
 138 my $xslt = XML::LibXSLT->new();
 139 my $parsed = $xslt->parse_stylesheet($style_doc);
 140
 141 my $biblio_html;
 142 my $biblio_parsed;
 143 my $biblio_data;
 144 my $biblio_author_external;
 145
 146 open(my $xml_fh, '>', '/tmp/bibliografija.xml') if $ENV{XML};
 147
 148 sub biblioitem_html {
 149         my ($biblionumber, $parse_only) = @_;
 150
 151         return $biblio_html->{$biblionumber} if exists $biblio_html->{$biblionumber} && ! $parse_only;
 152
 153         my $xmlrecord = $marcxml->{$biblionumber} || confess "missing $biblionumber marcxml";
 154
 155         print $xml_fh $xmlrecord if $ENV{XML};
 156
 157         my $source = eval { $parser->parse_string($xmlrecord) };
 158         if ( $@ ) {
 159 #               warn "SKIP $biblionumber corrupt XML";
 160                 push @{ $skip->{XML_corrupt} }, $biblionumber;
 161                 return;
 162         }
 163
 164         if ( $parse_only ) {
 165                 $biblio_parsed->{$biblionumber} = $source;
 166                 return $source;
 167         }
 168
 169         my $transformed = $parsed->transform($source);
 170         $biblio_html->{$biblionumber} = $parsed->output_string( $transformed );
 171
 172         delete $biblio_parsed->{$biblionumber};
 173
 174         return $biblio_html->{$biblionumber};
 175 }
 176
 177 $sth_select_authors->execute();
 178 while( my $row = $sth_select_authors->fetchrow_hashref ) {
 179 #       warn dump($row),$/;
 180
 181         my $biblio;
 182
 183         $marcxml->{ $row->{biblionumber} } = $row->{marcxml};
 184
 185         my $doc = biblioitem_html( $row->{biblionumber}, 1 );
 186         if ( ! $doc ) {
 187 #               warn "ERROR can't parse MARCXML ", $row->{biblionumber}, " ", $row->{marcxml}, "\n";
 188                 next;
 189         }
 190
 191         my $root = $doc->documentElement;
 192 =for leader
 193         my @leaders = $root->getElementsByLocalName('leader');
 194         if (@leaders) {
 195                 my $leader = $leaders[0]->textContent;
 196                 warn "leader $leader\n";
 197         }
 198 =cut
 199
 200         my $extract = {
 201                 '008' => undef,
 202                 '100' => '(9|a)',
 203                 '680' => 'i',
 204                 '700' => '(9|4|a)',
 205                 '942' => '(t|r|v)'
 206         };
 207
 208         my $data;
 209
 210         foreach my $elt ($root->getChildrenByLocalName('*')) {
 211                 my $tag = $elt->getAttribute('tag');
 212                 next if ! $tag;
 213                 next unless exists $extract->{ $tag };
 214
 215         if ($elt->localname eq 'controlfield') {
 216                         if ( $tag eq '008' ) {
 217                                 my $content = $elt->textContent;
 218                                 my $year = substr($content, 7, 4 );
 219                                 if ( $year !~ m/^\d+$/ ) {
 220                                         $year = 0;
 221                                         push @{ $skip->{invalid_year} }, $row->{biblionumber};
 222                                 }
 223                                 $biblio_year->{ $row->{biblionumber} } = $data->{year} = $year;
 224                                 $data->{'008'} = $content;
 225                         }
 226                         next;
 227         } elsif ($elt->localname eq 'datafield') {
 228                         my $sf_data;
 229             foreach my $sfelt ($elt->getChildrenByLocalName('subfield')) {
 230                 my $sf = $sfelt->getAttribute('code');
 231                                 next unless $sf =~ m/$extract->{$tag}/;
 232                                 if ( exists $sf_data->{$sf} ) {
 233                                         $sf_data->{$sf} .= " " . $sfelt->textContent();
 234                                 } else {
 235                                         $sf_data->{$sf} = $sfelt->textContent();
 236                                 }
 237                         }
 238                         push @{ $data->{$tag} }, $sf_data if $sf_data;
 239                 }
 240         }
 241
 242         if ( ! defined $data->{year} ) {
 243                 warn "MISSING year in ", $row->{biblionumber};
 244 =for remove-year-limit
 245         } elsif ( $data->{year} < 2008 ) {
 246                 push @{ $skip->{year_lt_2008} }, $row->{biblionumber};
 247                 next;
 248         } elsif ( $data->{year} > 2013 ) {
 249                 push @{ $skip->{year_gt_2013} }, $row->{biblionumber};
 250                 next;
 251 =cut
 252         }
 253
 254 #       warn "# ", $row->{biblionumber}, " data ",dump($data);
 255
 256         my $category = $data->{942}->[0]->{'t'};
 257         if ( ! $category ) {
 258 #               warn "# SKIP ", $row->{biblionumber}, " no category in ", dump($data);
 259                 push @{ $skip->{no_category} }, $row->{biblionumber};
 260                 next;
 261         }
 262
 263         my $have_100 = 1;
 264
 265         if ( exists $data->{100} ) {
 266                         my @first_author =
 267                                 map { $_->{'9'} }
 268                                 grep {
 269                                         if ( ! exists $_->{9} ) {
 270                                                 $biblio_author_external->{ $row->{biblionumber} }++;
 271                                                 0;
 272                                         } elsif ( exists $auth_header->{ $_->{9} } ) {
 273                                                 1; # from FFZXG
 274                                         } else {
 275                                                 0;
 276                                         }
 277                                 }
 278                                 @{ $data->{100} };
 279                         foreach my $authid ( @first_author ) {
 280                                 push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
 281                         }
 282         } else {
 283                 $have_100 = 0;
 284         }
 285
 286         my $have_edt;
 287
 288         if ( exists $data->{700} ) {
 289                         my @other_authors =
 290                                 grep {
 291                                         if ( ! exists $_->{9} ) {
 292                                                 $biblio_author_external->{ $row->{biblionumber} }++;
 293                                                 0;
 294                                         } elsif ( exists $auth_header->{ $_->{9} } ) {
 295                                                 1; # from FFZXG
 296                                         } else {
 297                                                 0;
 298                                         }
 299                                 }
 300                                 @{ $data->{700} };
 301                         foreach my $auth ( @other_authors ) {
 302                                 my $authid = $auth->{9} || next;
 303                                 my $type   = $auth->{4} || next; #die "no 4 in ",dump($data);
 304
 305                                 $type_stats->{$type}++;
 306
 307                                 if ( $type =~ m/(edt|trl|com|ctb)/ ) {
 308                                         push @{ $authors->{$authid}->{sec}->{ $category } }, $row->{biblionumber};
 309                                         push @{ $authors->{$authid}->{$1}->{ $category } }, $row->{biblionumber};
 310                                 } elsif ( $type =~ m/aut/ ) {
 311                                         if ( ! $have_100 ) {
 312                                                 $have_edt = grep { exists $_->{4} && $_->{4} =~ m/edt/ } @{ $data->{700} } if ! defined $have_edt;
 313                                                 if ( $have_edt ) {
 314                                                         $skip->{ have_700_edt }->{ $row->{biblionumber} }++;
 315                                                 } else {
 316                                                         push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
 317                                                 }
 318                                         } else {
 319                                                 push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
 320                                         }
 321                                 } else {
 322 #                                       warn "# SKIP ", $row->{biblionumber}, ' no 700$4 in ', dump($data);
 323                                         $skip->{ 'no_700$4' }->{ $row->{biblionumber} }++;
 324                                 }
 325                         }
 326                         delete $data->{700};
 327         }
 328
 329         $biblio_data->{ $row->{biblionumber} } = $data;
 330
 331 }
 332
 333 debug 'authors' => $authors;
 334 debug 'type_stats' => $type_stats;
 335 debug 'skip' => $skip;
 336 debug 'biblio_year' => $biblio_year;
 337 debug 'biblio_data' => $biblio_data;
 338 debug 'biblio_author_external' => $biblio_author_external;
 339
 340 my $category_label;
 341 my $sth_categories = $dbh->prepare(q{
 342 select authorised_value, lib from authorised_values where category = 'BIBCAT'
 343 });
 344 $sth_categories->execute();
 345 while( my $row = $sth_categories->fetchrow_hashref ) {
 346         $category_label->{ $row->{authorised_value} } = $row->{lib};
 347
 348 }
 349 debug 'category_label' => $category_label;
 350
 351 sub html_title {
 352         return qq|<html>
 353 <head>
 354 <meta charset="UTF-8">
 355 <title>|, join(" ", @_), qq|</title>
 356 <link href="style.css" type="text/css" rel="stylesheet" />
 357 <script src="//code.jquery.com/jquery-1.11.2.js"></script>
 358 </head>
 359 <body>
 360 |;
 361 }
 362
 363 sub html_end {
 364         return
 365                 qq|<small style="color:gray">Zadnji puta osvježeno: |,
 366                 strftime("%Y-%m-%d %H:%M:%S\n", localtime()),
 367                 qq|</body>\n</html>\n|;
 368 }
 369
 370 mkdir 'html' unless -d 'html';
 371
 372 open(my $index, '>:encoding(utf-8)', 'html/index.new');
 373 print $index html_title('Bibliografija Filozofskog fakulteta');
 374
 375 my $first_letter = '';
 376
 377 debug 'authors' => \@authors;
 378
 379 sub li_biblio {
 380         my ($biblionumber) = @_;
 381         return qq|<li class="y|, $biblio_year->{$biblionumber}, qq|">|,
 382                 qq|<a href="https://koha.ffzg.hr/cgi-bin/koha/opac-detail.pl?biblionumber=$biblionumber">$biblionumber</a>|,
 383                 biblioitem_html($biblionumber),
 384                 qq|<a href="https://koha.ffzg.hr:8443/cgi-bin/koha/cataloguing/addbiblio.pl?biblionumber=$biblionumber">edit</a>|,
 385                 qq|</li>\n|;
 386 }
 387
 388 sub unique_biblionumber {
 389         my @v = @_;
 390         my $u;
 391         $u->{$_}++ foreach @v;
 392         return sort { $biblio_year->{$b} <=> $biblio_year->{$a} || $a <=> $b } keys %$u;
 393 }
 394
 395 sub author_html {
 396         my ( $fh, $authid, $type, $label ) = @_;
 397
 398         return unless exists $authors->{$authid}->{$type};
 399
 400         print $fh qq|<h2>$label</h2>\n|;
 401
 402         foreach my $category ( sort keys %{ $authors->{$authid}->{$type} } ) {
 403                 my $label = $category_label->{$category} || 'Bez kategorije';
 404                 print $fh qq|<a name="$type-$category"><h3>$label</h3></a>\n<ul>\n|;
 405                 foreach my $biblionumber ( unique_biblionumber @{ $authors->{$authid}->{$type}->{$category} } ) {
 406                         print $fh li_biblio( $biblionumber );
 407                 }
 408                 print $fh qq|</ul>\n|;
 409         }
 410 }
 411
 412
 413 sub count_author_years {
 414         my $years = shift;
 415         my ($authid) = @_;
 416         foreach my $type ( keys %{ $authors->{$authid} } ) {
 417                 foreach my $category ( keys %{ $authors->{$authid}->{$type} } ) {
 418                         foreach my $biblionumber ( @{ $authors->{$authid}->{$type}->{$category} } ) {
 419                                 $years->{ $biblio_year->{ $biblionumber } }->{ $type . '-' . $category }++;
 420                         }
 421                 }
 422         }
 423         return $years;
 424 }
 425
 426 sub html_year_selection {
 427         my $fh = shift;
 428         my @authids = @_;
 429
 430         print $fh qq|<span id="years">Godine:\n|;
 431         my $type_cat_count = {};
 432         my $years;
 433
 434         foreach my $authid ( @authids ) {
 435                 $years = count_author_years( $years, $authid );
 436         }
 437
 438         foreach my $year ( sort { $b <=> $a } keys %$years ) {
 439                 print $fh qq|<label><input name="year_selection" value="$year" type=checkbox onClick="toggle_year($year, this)" checked="checked">$year</label>&nbsp;\n|;
 440                 foreach my $type_cat ( keys %{ $years->{$year} } ) {
 441                         $type_cat_count->{ $type_cat } += $years->{$year}->{$type_cat};
 442                 }
 443         }
 444
 445         print $fh qq|
 446 <input type=button value="all" onClick="all_years(1)">
 447 <input type=button value="none" onClick="all_years(0)">
 448         |;
 449
 450         print $fh qq|</span>|;
 451
 452         print $fh q|
 453 <script>
 454
 455 var years = |, encode_json($years), q|;
 456
 457 var type_cat_count = |, encode_json($type_cat_count), q|;
 458
 459 function year_show(year) {
 460         $('.y'+year).show();
 461         console.debug('show', year);
 462         for(var type_cat in years[year]) {
 463                 if ( ( type_cat_count[ type_cat ] += years[year][type_cat] ) == years[year][type_cat]) {
 464                         $('a[name="'+type_cat+'"]').show();
 465                         console.debug(type_cat, 'show');
 466                 }
 467         }
 468 }
 469
 470 function year_hide(year) {
 471         $('.y'+year).hide();
 472         console.debug('hide', year);
 473         for(var type_cat in years[year]) {
 474                 if ( ( type_cat_count[ type_cat ] -= years[year][type_cat] ) == 0 ) {
 475                         $('a[name="'+type_cat+'"]').hide();
 476                         console.debug(type_cat, 'hide');
 477                 }
 478         }
 479 }
 480
 481 function toggle_year(year, el) {
 482         if ( el.checked ) {
 483                 year_show(year);
 484         } else {
 485                 year_hide(year);
 486         }
 487 }
 488
 489 function all_years( turn_on ) {
 490         $('input[name=year_selection]').each( function(i,el) {
 491                 if ( turn_on ) {
 492                         if ( ! el.checked ) {
 493                                 el.checked = true;
 494                                 year_show( el.value );
 495                         }
 496                 } else {
 497                         if ( el.checked ) {
 498                                 el.checked = false;
 499                                 year_hide( el.value );
 500                         }
 501                 }
 502         } );
 503 }
 504
 505 $(document).ready( function() {
 506         console.info('ready');
 507
 508         $('input[name=year_selection]').each( function(i, el) {
 509                 var year = el.value;
 510                 console.debug( 'on load', year, el.checked );
 511                 if (! el.checked) year_hide(year);
 512         });
 513
 514 });
 515 </script>
 516         |;
 517 }
 518
 519 foreach my $row ( sort { $a->{full_name} cmp $b->{full_name} } @authors ) {
 520
 521         my $first = substr( $row->{full_name}, 0, 1 );
 522         if ( $first ne $first_letter ) {
 523                 print $index qq{</ul>\n} if $first_letter;
 524                 $first_letter = $first;
 525                 print $index qq{<h1>$first</h1>\n<ul>\n};
 526         }
 527         print $index qq{<li><a href="}, $row->{authid}, qq{.html">}, $row->{full_name}, "</a></li>\n";
 528
 529         my $path = "html/$row->{authid}";
 530         open(my $fh, '>:encoding(utf-8)', "$path.new");
 531         print $fh html_title($row->{full_name}, "bibliografija");
 532         print $fh qq|<h1>$row->{full_name} - bibliografija</h1>\n|;
 533
 534         html_year_selection $fh => $row->{authid};
 535
 536         author_html( $fh, $row->{authid}, 'aut' => 'Primarno autorstvo' );
 537         author_html( $fh, $row->{authid}, 'sec' => 'Uredništva, prijevodi, krička izdanja' );
 538
 539         print $fh html_end;
 540         close($fh);
 541         rename "$path.new", "$path.html";
 542
 543 }
 544
 545 print $index html_end;
 546 close($index);
 547 rename 'html/index.new', 'html/index.html';
 548
 549 debug 'auth_header' => $auth_header;
 550
 551
 552 my $department_category_author;
 553 foreach my $department ( sort keys %$auth_department ) {
 554         foreach my $authid ( sort @{ $auth_department->{$department} } ) {
 555                 my   @categories = keys %{ $authors->{$authid}->{aut} };
 556                 push @categories,  keys %{ $authors->{$authid}->{sec} };
 557                 foreach my $category ( sort @categories ) {
 558                         push @{ $department_category_author->{$department}->{$category} }, $authid;
 559                         push @{ $department_category_author->{'AAA_ukupno'}->{$category} }, $authid if $department_in_sum->{$department};
 560                         if ( my $group = $department_in_group->{ $department } ) {
 561                                 push @{ $department_category_author->{$group}->{$category} }, $authid;
 562                         } else {
 563                                 $skip->{'department_not_in_group'}->{ $department }++;
 564                         }
 565                 }
 566         }
 567 }
 568
 569 debug 'department_category_author' => $department_category_author;
 570
 571
 572 sub department_html {
 573         my ( $fh, $department, $type, $label ) = @_;
 574
 575         print $fh qq|<h2>$label</h2>\n|;
 576
 577         foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
 578
 579                 my @authids = @{ $department_category_author->{$department}->{$category} };
 580                 next unless @authids;
 581
 582                 my @biblionumber = unique_biblionumber map { @{ $authors->{$_}->{$type}->{$category} } } grep { exists $authors->{$_}->{$type}->{$category} } @authids;
 583
 584                 next unless @biblionumber;
 585
 586                 my $label = $category_label->{$category} || 'Bez kategorije';
 587                 print $fh qq|<a name="$type-$category"><h3>$label</h3></a>\n<ul>\n|;
 588
 589                 print $fh li_biblio( $_ ) foreach @biblionumber;
 590
 591                 print $fh qq|</ul>|;
 592         }
 593
 594 }
 595
 596
 597 mkdir 'html/departments' unless -d 'html/departments';
 598
 599 open(my $dep_fh, '>:encoding(utf-8)', 'html/departments/index.new');
 600 print $dep_fh html_title('Odsijeci Filozofskog fakulteta u Zagrebu'), qq|<ul>\n|;
 601 foreach my $department ( sort keys %$department_category_author ) {
 602         my $dep = $department || 'Nema odsjeka';
 603         my $dep_file = unac_string('utf-8',$dep);
 604         print $dep_fh qq|<li><a href="$dep_file.html">$dep</a></li>\n|;
 605         open(my $fh, '>:encoding(utf-8)', "html/departments/$dep_file.new");
 606
 607         print $fh html_title($department . ' bibliografija');
 608         print $fh qq|<h1>$department bibliografija</h1>\n|;
 609
 610         my @authids;
 611         foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
 612                 push @authids, @{ $department_category_author->{$department}->{$category} };
 613         }
 614         html_year_selection $fh => @authids;
 615
 616         department_html( $fh, $department, 'aut', 'Primarno autorstvo' );
 617
 618         department_html( $fh, $department, 'sec', 'Sekundarno autorstvo' );
 619
 620         print $fh html_end;
 621         close($fh);
 622         rename "html/departments/$dep_file.new", "html/departments/$dep_file.html";
 623 }
 624 print $dep_fh qq|</ul>\n|, html_end;
 625 close($dep_fh);
 626 rename 'html/departments/index.new', 'html/departments/index.html';
 627
 628 my $azvo_stat_biblio;
 629
 630 foreach my $department ( sort keys %$department_category_author ) {
 631         foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
 632                 foreach my $authid ( @{ $department_category_author->{$department}->{$category} } ) {
 633                         my $group = $auth_group->{$authid};
 634                         if ( ! $group ) {
 635                                 push @{ $skip->{no_auth_group} }, $authid;
 636                                 next;
 637                         }
 638                         foreach my $type ( keys %{ $authors->{$authid} } ) {
 639                                 next unless exists $authors->{$authid}->{$type}->{$category};
 640                                 push @{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{$group} },  @{ $authors->{$authid}->{$type}->{$category} };
 641                                 push @{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{''} },  @{ $authors->{$authid}->{$type}->{$category} };
 642                         }
 643                 }
 644                 foreach my $type ( keys %{ $azvo_stat_biblio->{ $department }->{ $category } } ) {
 645                         foreach my $group ( keys %{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type } } ) {
 646                                 my @biblios = unique_biblionumber @{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{ $group } };
 647                                 $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{ $group } = [ @biblios ];
 648                         }
 649                 }
 650         }
 651 }
 652
 653 debug 'azvo_stat_biblio' => $azvo_stat_biblio;
 654
 655 my @report_lines;
 656 my @report_labels;
 657
 658 my $label;
 659 my $sub_labels;
 660 open(my $report, '<:encoding(utf-8)', 'AZVO.txt');
 661 while( <$report> ) {
 662         chomp;
 663         if ( /^([^\t]+)\t+(.+)/ ) {
 664                 $label = $1;
 665                 push @report_labels, $label;
 666                 my $type = [ map { m/\s+/ ? [ split(/\s+/,$_) ] : [ $_, 'aut' ] } split (/\s*\+\s*/, $2) ];
 667                 push @report_lines, [ $label, @$type ];
 668         } elsif ( /^\t+([^\t]+):\t+(\d+)(\w*)\t*(.*)$/ ) {
 669                 push @{ $sub_labels->{$label} }, [ $1, $2, $3, $4 ];
 670                 my $sub_label = $1;
 671                 pop (@report_labels) if ( $report_labels[ $#report_labels ] =~ m/^$label$/ ); # remove partial name
 672                 push @report_labels, $label . $sub_label;
 673         } else {
 674                 die "ERROR: [$_]\n";
 675         }
 676 }
 677
 678 debug 'report_lines', \@report_lines;
 679 debug 'sub_labels', $sub_labels;
 680 debug 'report_labels', \@report_labels;
 681
 682 my @departments = ( sort { lc($a) cmp lc($b) } keys %$azvo_stat_biblio );
 683
 684 debug 'departments' => \@departments;
 685
 686 my $department2col;
 687 $department2col->{ $departments[$_] } = $_ foreach ( 0 .. $#departments );
 688 my $label2row;
 689 $label2row->{ $report_labels[$_] } = $_ foreach ( 0 .. $#report_labels );
 690
 691 my $table;
 692
 693 sub table_count {
 694         my $label = shift @_;
 695         my $department = shift @_;
 696         my $group = shift @_;
 697         my @biblionumbers = @_;
 698         my $unique;
 699         $unique->{$_}++ foreach @biblionumbers;
 700         my @bibs = keys %$unique;
 701         $table->{ffzg}->{$group}->[ $label2row->{ $label } ]->[ $department2col->{$department} ] = scalar @bibs;
 702         $table->{external}->{$group}->[ $label2row->{ $label } ]->[ $department2col->{$department} ] = scalar grep { $biblio_author_external->{$_} } @bibs;
 703 }
 704
 705 foreach my $group ( '', keys %$azvo_group_title ) {
 706
 707 foreach my $department ( @departments ) {
 708         foreach my $line ( @report_lines ) {
 709                 my $label = $line->[0];
 710                 my @biblionumbers;
 711                 foreach ( 1 .. $#$line ) {
 712                         my ( $category, $type ) = @{ $line->[ $_ ] };
 713                         my $b = $azvo_stat_biblio->{ $department }->{$category}->{$type}->{$group};
 714                         push @biblionumbers, @$b if $b;
 715                 }
 716                 if ( $sub_labels->{$label} ) {
 717                         my $sub_stats;
 718                         foreach my $biblionumber ( @biblionumbers ) {
 719                                 my $data = $biblio_data->{$biblionumber} || die "can't find biblionumber $biblionumber";
 720                                 foreach my $sub_label ( @{ $sub_labels->{$label} } ) {
 721                                         my ( $sub_label, $field, $sf, $regex ) = @$sub_label;
 722                                         if ( ! $regex ) {
 723                                                 push @{ $sub_stats->{ $sub_label } }, $biblionumber;
 724                                                 last;
 725                                         }
 726                                         if ( $field < 100 ) {
 727                                                 if ( $data->{$field} =~ m/$regex/ ) {
 728                                                         push @{ $sub_stats->{ $sub_label } }, $biblionumber;
 729                                                         last;
 730                                                 }
 731                                         } else {
 732                                                 if ( exists $data->{$field}->[0]->{$sf} && $data->{$field}->[0]->{$sf} =~ m/$regex/ ) {
 733                                                         push @{ $sub_stats->{ $sub_label } }, $biblionumber;
 734                                                         last;
 735                                                 }
 736                                         }
 737                                 }
 738                         }
 739                         foreach my $sub_label ( keys %$sub_stats ) {
 740                                 my $full_label = $label . $sub_label;
 741                                 table_count $full_label, $department, $group, @{ $sub_stats->{$sub_label} };
 742                         }
 743                 } else {
 744                         table_count $label, $department, $group, @biblionumbers;
 745                 }
 746         }
 747 }
 748
 749 } # group
 750
 751 debug 'table', $table;
 752
 753 open(my $fh, '>:encoding(utf-8)', 'html/azvo.new');
 754 open(my $fh2, '>:encoding(utf-8)', 'html/azvo2.new');
 755
 756 sub print_fh {
 757         print $fh @_;
 758         print $fh2 @_;
 759 }
 760
 761 print $fh html_title('AZVO tablica - FFZG');
 762 print $fh2 html_title('AZVO tablica - kolaboracija sa FFZG');
 763
 764 foreach my $group ( keys %{ $table->{ffzg} } ) {
 765
 766                 print_fh "<h1>$group</h1>" if $group;
 767
 768                 print_fh "<table border=1>\n";
 769                 print_fh "<tr><th></th>";
 770                 print_fh "<th>$_</th>" foreach @departments;
 771                 print_fh "</tr>\n";
 772
 773                 foreach my $row ( 0 .. $#{ $table->{ffzg}->{$group} } ) {
 774                         print_fh "<tr><th>", $report_labels[$row], "</th>\n";
 775                         foreach ( 0 .. $#departments ) {
 776                                 print_fh "<td>";
 777                                 print $fh $table->{ffzg}->{$group}->[ $row ]->[ $_ ] || '';
 778                                 print $fh2 $table->{external}->{$group}->[ $row ]->[ $_ ] || '';
 779                                 print_fh "</td>\n"
 780                         }
 781                         print_fh "</tr>\n";
 782                 }
 783
 784                 print_fh "</table>\n";
 785
 786 } # group
 787
 788 print_fh html_end;
 789 close($fh);
 790 close($fh2);
 791 rename 'html/azvo.new', 'html/azvo.html';
 792 rename 'html/azvo2.new', 'html/azvo2.html';
 793